PROJECT ABSTRACT Major depressive disorder (MDD), anxiety disorders, and substance use disorders (SUDs) are common, complex psychiatric traits that frequently co-occur and are associated with significant functional impairment, increased healthcare utilization and cost, and higher mortality risk. Not only are these three conditions highly prevalent in the general population and generate a huge societal burden, but recent studies by our team and others have shown that shared covariance from common genetic variation significantly contributes to these psychiatric comorbidities. Large data sets are needed to understand how the multifaceted interplay of genetics, including polygenic risk scores (PRSs), and social determinants of health factors, such as employment and educational attainment, can increase the risk of these psychiatric disorders and clinical outcomes, such as multiple psychiatric hospitalizations. PRSs have shown potential for risk prediction, but the clinical utility of PRSs for psychiatric conditions is just starting to be explored. Use of Electronic Health Records (EHRs) offers the promise of large data sets to examine these relationships in cohorts of patients seen in clinical practice. However, the use of EHRs is in its infancy in the study of psychiatric disorders and their treatment. This study will address critical knowledge gaps in ?genotype-psychiatric phenotype? relationships in large, demographically and geographically diverse population-based samples derived from EHR-linked biobanks across four medical centers - Columbia, Cornell, Mayo Clinic and Mount Sinai. Our objectives are to (1) develop improved methods for EHR phenotyping of MDD, anxiety, and SUDs, and related outcomes based on a data-set of >30 million EHRs, (2) evaluate associations between PRSs and these conditions, as well as (3) assess the association between PRSs and outcomes including treatment resistance in MDD and healthcare utilization in patients with MDD, anxiety and SUD. The PRS analyses will utilize data from biobanks with >50,000 persons with both EHR and GWAS data. Successful completion of this study will generate new data in improving our understanding of the clinical utility of PRSs for commonly occurring psychiatric disorders.